Introduction

As the world’s population continues to increase, the energy required to power it increases as well [1]. To keep up with the ever-growing energy demand, larger amounts of fossil fuels are being used each year causing the global emissions of CO2 to increase at an unsustainable rate [2, 3]. For the world to maintain the current energy demand while also reducing CO2 emissions, the development of carbon–neutral energy sources is crucial. Solid oxide fuel cells (SOFCs) are a promising source of carbon–neutral energy due to their high chemical-to-electrical energy conversion process that does not produce the harmful pollutants (when hydrogen is used as a fuel) associated with combustion reactions [4]. Apart from hydrogen, SOFCs can also use other fuels such as methane, carbon monoxide, and other hydrocarbons due to the higher operating temperature compared to other fuel cells, including proton exchange membrane fuel cells and direct methanol fuel cells [5]. SOFCs commonly operate at a high temperature ranging from 800 to 1000 °C. However, at these temperatures, long-term cell degradation is an issue as the components that are in contact with each other will react and form less conductive products [6]. Thus, it is important to lower the operating temperature of these cells to an intermediate temperature (IT) range (400–800 °C) to increase their lifespan and limit the drop in efficiency over time. However, at these lower temperatures, the kinetics for the oxygen reduction (ORR) at the cathode are slow; therefore, advanced cathodes with high catalytic activity for the ORR must be developed [7, 8].

Recent studies have focused on improving the performance of various mixed ionic-electronic conducting metal oxide perovskites for SOFCs. Extensive research has been done on the optimization of La0.6Sr0.4Co0.2Fe0.8O3-δ (LSCF) cathodes due to their high total electrical conductivity and high electrochemical activity towards the ORR. However, LSCF has shown durability issues mainly caused by electrode poisoning leading to degradation of the LSCF cathode [9]. LSCF has also been shown to react with the La0.8Sr0.2Ga0.8Mg0.2O3-δ (LSGM) electrolyte which lowers the performance of the cell [10]. Our group found better compatibility with the LSGM electrolyte by replacing La in LSCF with Nd and changing the stoichiometry to Nd0.75Sr0.25Co0.8Fe0.2O3-δ (NSCF). The better compatibility is attributed to a lower thermal expansion coefficient (TEC) that is closer to the TEC of LSGM and no reactivity between NSCF and LSGM [11]. Substituting an A-site ion for one with a larger ionic radius increases the free lattice volume allowing for greater oxygen diffusion and transport through oxygen vacancies. Thus, we were curious about the effect of replacing Sr with Ba in Nd0.75Sr0.25Co0.8Fe0.2O3−δ [12]. A similar NdBaCoFeO5+δ double perovskite has been studied previously, but the single perovskite variation has not been looked at so far [13]. In this paper, a new Nd-based Nd0.75Ba0.25Co0.8Fe0.2O3-δ (NBCF25) perovskite is synthesized using a high temperature, solid-state reaction, and the phase formation, electrical conductivity from 100 to 800 °C, ASR from 650 to 800 °C, and the rate-limiting step of the ORR from 650 to 800 °C is determined.

Experimental aspects

Material synthesis

The perovskite-type Nd0.75Ba0.25Co0.8Fe0.2O3-δ was prepared by a solid-state reaction in air at an elevated temperature. Neodymium oxide (Nd2O3) (99.9% Sigma-Aldrich), barium carbonate (BaCO3) (99.9% Alfa Aesar), cobalt oxide (Co3O4) (99% Alfa Aesar), and iron oxide (Fe2O3) (99 + % Sigma-Aldrich) were mixed in required stoichiometric quantities. The precursors were ball milled with isopropanol for 3 h followed by calcination at 900 °C for 15 h in air. The calcined samples were ball milled again for 3 h to reduce the particle size and then pressed into circular pellets (diameter of ≈ 1.2 cm and length of ≈ 1 cm). The pellets were sintered at 1300 °C for 10 h in air. Finally, the pellets were crushed, and ball milled for 3 h followed by sintering to give a fine powder. La0.8Sr0.2Ga0.8Mg0.2O3-δ (LSGM) was obtained from fuelcellmaterials, a Nexceris company, USA, and pressed into circular pellets (≈ 0.5 mm in thickness and ≈ 10 mm in diameter) and sintered at 1350 °C for 8 h.

Phase characterization

Powder X-ray diffraction (PXRD) measurements of the as-prepared samples were obtained using a Rigaku SmartLab® diffractometer (CuKα radiation, λKα1 = 1.540593 Å, λKα2 = 1.544414 Å; 40 kV and 50 mA; 0.01° steps, 0.5°/min). The powder samples were packed into an aluminum well with a depth of 0.3 mm to increase the signal-to-noise ratio. An X-ray fluorescence reduction mode was used on the detector (Rigaku HYPIX-3000 detector in 1-D mode). The incident slit was set to 0.5° with an incident mask of 10 mm. The receiving optics was set at 20 mm. A 2.5° solar slit was used on both the incident and receiving sides. A CuKβ filter was used to remove the Kβ radiation, and an X-ray anti-scattering device was also used to limit background radiation. The samples were spun at 60 rpm for averaging data.

4-probe DC chronopotentiometry measurements

4-probe DC chronopotentiometry measurements were performed on the sintered sample pellets (diameter of ≈ 1.2 cm and length of ≈ 1 cm) using a BioLogic VSP-300 potentiostat. Groves were cut into the sample pellet and Au wires were wrapped around these grooves to create the inner probes where the potential is measured. The sides of the pellets were coated with gold which served as a current collector. A constant current (0.1–0.35 A) was applied on the sides and the resulting voltage was measured. The following equation was used to calculate the conductivity where I is current (A), V is potential (V), L is the length between the inner probes (cm), and A is the area of the outer probes [14] (cm2):

$$\sigma =\frac{I}{V}\times \frac{L}{A}$$
(1)

Microstructural analysis

The microstructure was measured by scanning electron microscopy (SEM) with a field emission gun (ZEISS Sigma) operating at an accelerated voltage of 10 kV, a working distance of 6.7 mm, and 104 times magnification.

Electrochemical impedance spectroscopy measurements (EIS)

A VersaSTAT 3 potentiostat and galvanostat was used to perform the EIS measurements using alternating current (AC) with an amplitude of 10 mV over a frequency range of 10−1–106 Hz in air. These symmetric cells were prepared by screen printing a composite slurry (70 wt.% NBCF25 and 30 wt.% LSGM) onto an LSGM pellet (≈ 0.5 mm in thickness and ≈ 10 mm in diameter) and then sintering the pellet at 1050 °C for 5 h. These pellets were coated with gold which served as a current collector.

Impedance spectroscopy genetic programming (ISGP)

ISGP was used to generate a distribution function of relaxation times (DFRT), a model that represents different processes as peaks in the time domain [15,16,17]. ISGP utilizes the following equation to generate a Nyquist plot based on the DFRT that best fits the measured [18] impedance data:

$$Z\left(\omega \right)={R}_{\infty }+{R}_{pol}{\int }_{-\infty }^{\infty }\frac{\Gamma (\mathrm{log}(\tau ))}{1+i\omega \tau }d(\mathrm{log}(\tau ))$$
(2)

where Z is the impedance, R∞ is the series resistance, Rpol is the total polarization resistance, Γ is the DFRT, τ is the relaxation time, and ω is the angular frequency. The ISGP program will generate several DFRT models and grade them based on a “fitness function” based on its compatibility with the experimental impedance data; the model with the highest grade will be used for generation 1. The next generation will be an evolution of the previous generation with minor changes to the DFRT (“mutation”) and these new models are again ranked and the model with the highest grade is used to produce the next generation. The program reaches its final model when all mutations of the current model result in a lower grade [19].

Results and discussion

Phase characterization

Powder X-ray diffraction refinement of NBCF25 shows the formation of a single-phase orthorhombic phase in the Pnma space group (Fig. 1). The quality of the refinement is supported by a weighted reliability factor (Rwp) of 7.3% and a chi-squared (χ2) of 1.13. The A-site occupies the 4c Wyckoff position and is made up of 75% Nd and 25% Ba. The B-site occupies the 4b Wyckoff positive and is made up of 80% Co and 20% Fe. There are two types of oxygen sites one occupying the 4c Wyckoff position and the other occupying the 8d Wyckoff position. Figure 2 shows the crystal structure of Nd0.75Ba0.25Co0.8Fe0.2O3-δ.

Fig. 1
figure 1

PXRD Rietveld refinement profile for Nd0.75Ba0.25Co0.8Fe0.2O3-δ. Details of refinement results are provided in. Table 1 [20]

Full size image
Table 1 The powder X-ray refinement results for Nd0.75Ba0.25Co0.8Fe0.2O3-δ (Rwp = 7.343%; χ2 = 1.13; phase Nd0.75Ba0.25Co0.8Fe0.2O3-δ: space group = Pnma, a = 5.4160(9) Å, b = 7.6642(5) Å, c = 5.4489(8) Å). Occupancy parameters are not refined
Full size table
Fig. 2
figure 2

The crystal structure of Nd0.75Ba0.25Co0.8Fe0.2O3-δ determined using Rietveld refinement. 3D rendering generated by VESTA 3. [21] A-site consists of 75% Nd and 25% Ba. B-site consists of 80% Co and 20% Fe

Full size image

DC electrical conductivity

The electrical conductivity as a function of temperature as well as the TGA curve for NBCF25 is shown in Fig. 3. NBCF25 is a p-type semiconductor. The conductivity trends can be broken down into 5 different regions. In the first region from 100 to 300 °C, there is an increase in conductivity as temperature increases; there is also a slight weight loss associated desorption of atmospheric H2O and CO2. In the second region from 300 to 400 °C, there is a sudden drop in conductivity and a sudden weight loss associated with the reduction of the B-site ions (Fe4+/Co4+) and the loss of oxygen as shown in the forward reaction of Eq. 3; this suggests the formation of \({\mathrm{B}}_{\mathrm{B}}^{\mathrm{X}}-{\mathrm{O}}_{\mathrm{O}}^{\mathrm{X}}-{\mathrm{B}}_{\mathrm{B}}^{\mathrm{X}}\) and \({\mathrm{B}}_{\mathrm{B}}^{\mathrm{X}}-{\mathrm{V}}_{\mathrm{O}}^{\bullet \bullet }-{\mathrm{B}}_{\mathrm{B}}^{\bullet }\) which block the transport of electrons where \({\mathrm{B}}_{\mathrm{B}}^{\mathrm{X}}\) is Fe3+ or Co3+ sitting at the Fe3+ or Co3+ site respectively, \({\mathrm{O}}_{\mathrm{O}}^{\mathrm{X}}\) is oxygen sitting at the oxygen site, \({\mathrm{V}}_{\mathrm{O}}^{\bullet \bullet }\) is a vacancy at the O2− site, and \({\mathrm{B}}_{\mathrm{B}}^{\bullet }\) is Fe4+ or Co4+ sitting at the Fe3+ or Co3+ site respectively. [22] In the third region from 400 to 500 °C, there is a sudden increase in conductivity and a sudden weight gain which is associated with the oxidation of the previously reduced B-site ions (Fe3+/Co3+) and the re-absorption of oxygen as shown in the reverse reaction of Eq. 2. In the fourth region from 500 to 700 °C, there is a rapid loss of weight but unlike region 2, the conductivity increases indicating semiconducting behavior. The fifth region from 700 to 800 °C has the same trend as the second region.

Fig. 3
figure 3

Arrhenius plot of the electrical conductivity for Nd0.75Ba0.25Co0.8Fe0.2O3-δ in air (blue) and TGA in N2 (red)

Full size image
$$2{\mathrm{B}}_{\mathrm{B}}^{\bullet }+{\mathrm{O}}_{\mathrm{O}}^{\mathrm{X}}\leftrightarrows 2{\mathrm{B}}_{\mathrm{B}}^{\mathrm{X}}+{\mathrm{V}}_{\mathrm{O}}^{\bullet \bullet }+\frac{1}{2}{\mathrm{O}}_{2}$$
(3)

Chemical reactivity with LSGM

Figure 4 shows the PXRD measurements for the LSGM electrolyte, the as-prepared NBCF25, 70% NBCF25 mixed with 30% LSGM, and 70% NBCF25 sintered with 30% LSGM at 1050 °C. A new peak arises at 2θ = 30.08° which indicates the reaction between NBCF25 and LSGM forming BaLaGa3O7 or BaNdGa3O7 [23]. Despite the reactivity, the area specific resistance of a LSGM-NBCF25 symmetric cell is low (Fig. 5) and SEM image (Fig. 6) of the cross-section shows good contact between LSGM and NBCF25 indicating that the formation of BaLaGa3O7 or BaNdGa3O7 has little effect on the performance of NBCF25 when used with LSGM.

Fig. 4
figure 4

PXRD of La0.8Sr0.2Ga0.8Mg0.2O3-δ (LSGM); Nd0.75Ba0.25Co0.8Fe0.2O3-δ (NBCF25), a composite of 70% NBCF25 and 30% LSGM, and a composite of 70% NBCF25 and 30% LSGM sintered at 1050 °C for 5 h. BaLaGa3O7 or BaNdGa3O7 (*)

Full size image
Fig. 5
figure 5

A Nyquist plot of a symmetrical cell with a composite electrode (70% Nd0.75Ba0.25Co0.8Fe0.2O3-δ + 30% LSGM) using LSGM with a thickness of 0.5 mm as an electrolyte where the ASR is equal to half the diameter of the arcs. The solid lines on the graph are generated from the equivalent circuit model given by Eq. 2. Measurements are taken at 650 °C to 800 °C in air. Corrected for cathode area. Inductance and series resistance is removed. B DFRT model of a Nyquist plot for Nd0.75Ba0.25Co0.8Fe0.2O3-δ where each peak corresponds to an arc on the Nyquist plot. C Linear Kramers–Kronig validity test performed on Nyquist data at 650 to 800 °C. The high-frequency inductance was removed at all temperatures

Full size image
Fig. 6
figure 6

Cross-sectional SEM images of Nd0.75Ba0.25Co0.8Fe0.2O3-δ (NBCF25)/La0.8Sr0.2Ga0.8Mg0.2O3-δ (LSGM) composite cathode screen printed onto LSGM electrolyte

Full size image

Area specific resistance (ASR)

Figure 6 shows a cross-sectional SEM image of an NBCF25 symmetric cell with a porous cathode layer and a dense LSGM electrolyte. The strong contact between NBCF25 and LSGM indicates good surface compatibility when sintered at 1050 °C. The area-specific resistance of a symmetric NBCF25 cell at various intermediate temperatures (650 °C to 800 °C) is shown in Fig. 5A. The ASR values are 0.49, 0.2, 0.11, and 0.07 Ω cm2 at 650 °C, 700 °C, 750 °C, and 800 °C respectively. Thus, NBCF25 has high catalytic activity towards the ORR at 750 °C and 800 °C which is determined through its low ASR at these temperatures (≈ 0.1 Ω cm2). Figure 5B shows the DFRT plot for NBCF25 with the inductance and series resistance removed where each peak corresponds to the frequency (f) of an individual electrical process which mimics a resistor (R) and a constant phase element (CPE) in parrel, known as an RC element. The area under each curve in the DFRT plot corresponds to the percentage of the total ASR that is due to the resistance of that electrical process. Then, as the frequency and resistance of the RC element are known, the following formula can be used to calculate [19] capacitance (C):

$$C=\frac1{2\pi Rf}$$
(4)

which is related to CPE [24]. At 650, 700, and 800 °C, there are two medium-frequency peaks associated with the coupled dissociative adsorption or surface exchange of oxygen and one low-frequency peak associated with oxygen diffusion inside the cathode [25, 26]. At 650 °C, the total resistance is dominated by oxygen diffusion inside the cathode (≈ 91%) (Table 2). At 700 °C, the total resistance becomes dominated by the coupled dissociative adsorption or surface exchange of oxygen (≈ 70%). At 800 °C, the coupled dissociative adsorption or surface exchange of oxygen and oxygen diffusion inside the cathode contribute equally to the total resistance. At 750 °C, there is one high-frequency peak associated with ion transfer from the cathode to the electrolyte, one medium-frequency, and one low-frequency peak [26]. The total resistance at 750 °C is dominated by coupled dissociative adsorption or surface exchange of oxygen. While contributing very little to the total resistance at 750 °C, the high-frequency charge transfer peak indicates minor instability between NBCF25 and LSGM at this temperature.

Table 2 Calculated capacity and resistance of all RC elements from DFRT of NBCF25
Full size table

The quality of the data used for the ISGP fitting was confirmed using a linear Kramers–Kronig validity test (Fig. 5C) [27,28,29] (http://www.Iwe.Kit.Edu/Lin-KK.Php). All data points on the Z′ and Z″ axis are within 0.1% of their expected value according to the formula [27]:

$${\Delta }_{{Z}^{^{\prime}}}=\frac{{Z}^{^{\prime}}-{Z}_{KK}^{^{\prime}}}{{Z}_{KK}^{^{\prime}}} \mathrm{or} {\Delta }_{{Z}^{^{\prime\prime} }}=\frac{{Z}^{^{\prime\prime} }-{Z}_{KK}^{^{\prime\prime} }}{{Z}_{KK}^{^{\prime\prime} }}$$
(5)

where ΔZ′ and ΔZ″ are the residuals for the real and imaginary axis respectively, Z′ and Z″ are the experimentally determined real and imaginary values respectively, and Z′KK and Z″KK are the expected real and imaginary values respectively determined from the Kramers–Kronig relations. The residuals all being near or less than 0.1% indicate the data is of very high quality [30, 31].

Table 3 shows a comparison between NBCF25 and various literature perovskite and double perovskite-type cathode materials for IT-SOFCs. NBCF25 has a comparable ASR to NdBaCoFeO5+δ and La0.6Sr0.4Co0.2Fe0.8O3-δ (≈ 0.1 Ω cm2 at 750 °C) [13, 32]. It has a very high electrical conductivity when compared to other perovskites (674 S/cm at 700 °C), 9 times more conductive than NdBaCoFeO5+δ and 3 times more conductive than La0.6Sr0.4Co0.2Fe0.8O3-δ [13, 33].

Table 3 Summary of the conductivity and ASR of NBCF and various literature perovskite and double perovskite-type cathode materials for IT-SOFCs
Full size table

Conclusions

In summary, perovskite-type Nd0.75Ba0.25Co0.8Fe0.2O3-δ (NBCF25) was prepared via solid-state synthesis method at 1300 °C in air. Powder X-ray diffraction showed the formation of a single-phase Nd0.75Ba0.25Co0.8Fe0.2O3-δ phase in the Pnma space group. NBCF25 shows minor reactivity with SrO/MgO-doped lanthanum gallate (LSGM) when sintered with 30% LSGM at 1050 °C. The conductivity of NBCF25 was determined using 4-probe DC chronopotentiometry on the sintered sample pellets and showed a maximum conductivity of 674 S/cm at 700 °C. The area-specific resistance (ASR) was determined using 2-probe AC electrochemical impedance spectroscopy on LSGM pellets with a screen-printed composite slurry (70 wt.% NBCF25 and 30 wt.% LSGM) and showed a minimum ASR of 0.07 Ω cm2 at 800 °C.